Background: Venous thromboembolism (VTE) is a severe complication in critically ill patients, often resulting in death and long-term disability and is one of the major contributors to the global burden of disease. This study aimed to construct an interpretable machine learning (ML) model for predicting VTE in critically ill patients based on clinical features and laboratory indicators. Methods: Data for this study were extracted from the eICU Collaborative Research Database (version 2.0). A stepwise logistic regression model was used to select the predictors that were eventually included in the model. The random forest, extreme gradient boosting (XGBoost) and support vector machine algorithms were used to construct the model using fivefold cross-validation. The area under curve (AUC), accuracy, no information rate, balanced accuracy, kappa, sensitivity, specificity, precision, and F1 score were used to assess the model's performance. In addition, the DALEX package was used to improve the interpretability of the final model. Results: This study ultimately included 109,044 patients, of which 1647 (1.5%) had VTE during ICU hospitalization. Among the three models, the Random Forest model (AUC: 0.9378; Accuracy: 0.9958; Kappa: 0.8371; Precision: 0.9095; F1 score: 0.8393; Sensitivity: 0.7791; Specificity: 0.9989) performed the best. Conclusion: ML models can be a reliable tool for predicting VTE in critically ill patients. Among all the models we had constructed, the random forest model was the most effective model that helps the user identify patients at high risk of VTE early so that early intervention can be implemented to reduce the burden of VTE on the patients.

Background: Breastfeeding is essential for maintaining the health of mothers and babies. Breastfeeding can reduce the infection rate and mortality in newborns, and can reduce the chances of overweight and obesity in children and adolescents. For mothers, a longer duration of breastfeeding can reduce the risk of breast cancer, ovarian cancer, and type 2 diabetes. Although breastfeeding has many benefits, the global breastfeeding rate is low. With the progress of time, the popularity of mobile devices has increased rapidly, and interventions based on mobile health (mHealth) may have the potential to facilitate the improvement of the breastfeeding status. Objective: The main objective of this study was to analyze the existing evidence to determine whether mHealth-based interventions can improve the status of breastfeeding. Methods: We systematically searched multiple electronic databases (PubMed, Web of Science, The Cochrane Library, Embase, CNKI, WanFang, and Vip ) to identify eligible studies published from 1966 to October 29, 2020. Included studies were randomized controlled trials (RCTs) studying the influence of mHealth on breastfeeding. The Cochrane Collaboration Risk of Bias tool was used to examine the risk of publication bias. RevMan 5.3 was used to analyze the data. Results: A total of 15 RCTs with a total sample size of 4366 participates met the inclusion criteria. Compared with usual care, interventions based on mHealth significantly increased the postpartum exclusive breastfeeding rate (odds ratio [OR] 3.18, 95% CI 2.20-4.59; P<.001), enhanced breastfeeding self-efficacy (mean difference [MD] 8.15, 95% CI 3.79-12.51; P=.002; I2=88%), reduced health problems in infants (OR 0.62, 95% CI 0.43-0.90; P=.01; I2=0%), and improved participants’ attitudes toward breastfeeding compared with usual care (MD 3.94, 95% CI 1.95-5.92; P<.001; I2=0%). There was no significant difference in the initiation of breastfeeding within an hour of birth between the intervention group and the usual care group (OR 1.26, 95% CI 0.55-2.90; P=.59). In addition, subgroup analysis was carried out according to different subjects and publication times. The results showed that the breastfeeding rate was not limited by the types of subjects. The breastfeeding rate based on mHealth at 1 month and 2 months after delivery did not change over the time of publication (2009 to 2020), and the breastfeeding rate based on mHealth at 3 months and 6 months after delivery gradually increased with time (2009 to 2020). Conclusions: Interventions based on mHealth can significantly improve the rate of postpartum exclusive breastfeeding, breastfeeding efficacy, and participants’ attitudes toward breastfeeding, and reduce health problems in infants. Therefore, encouraging women to join the mHealth team is feasible, and breastfeeding-related information can be provided through simple measures, such as text messages, phone calls, and the internet, to improve the health of postpartum women and their babies.

Objectives An up-to-date, detailed global analysis of the current status of the metabolic-attributed cardiovascular disease (CVD) burden has not been reported. Therefore, we investigated the global burden of metabolic-attributed CVD and its association with socioeconomic development status over the past 30 years. Methods Data on the burden of metabolic-attributed CVD were taken from the 2019 Global Burden of Disease (GBD) study. Metabolic risk factors of CVD included high fasting plasma glucose, high low-density lipoprotein cholesterol (LDL-c), high systolic blood pressure (SBP), high body mass index (BMI) and kidney dysfunction. Numbers and age-standardised rates (ASR) of disability-adjusted life-years (DALYs) and deaths were extracted and stratified by sex, age, Socio-demographic Index (SDI) level, country and region. Results The ASR of metabolic-attributed CVD DALYs and deaths decreased by 28.0% (95% UI 23.8% to 32.5%) and 30.4% (95% UI 26.6% to 34.5%), respectively, from 1990 to 2019. The highest burden of metabolic-attributed total CVD and intracerebral haemorrhage was mainly in low SDI locations, while the highest burden of ischaemic heart disease and IS was mainly in high SDI locations. The burden of DALYs and deaths in CVD was higher in men than women. In addition, the number and ASR of DALYs and deaths were highest in those over 80 years old. Conclusion Metabolic-attributed CVD threatens public health, especially in low-SDI locations and among the elderly. Low SDI location should strengthen the control of metabolic factors such as high SBP, high BMI, and high LDL-c and increase the knowledge of metabolic risk factors for CVD. Countries and regions should enhance screening and prevention of metabolic risk factors of CVD in the elderly. Policy-makers should use 2019 GBD data to guide cost-effective interventions and resource allocation.

Aims: Rheumatic heart disease (RHD) remains an important health issue, yet global attention to RHD is diminishing. This study aimed to investigate the global burden of RHD and its relationship with socioeconomic development status. Methods and results: Data were obtained from the Global Burden of Disease (GBD) 2019 database. Incidence, prevalence, disability-adjusted life years (DALYs), and mortality numbers and rates for RHD were extracted and stratified by sex, level of socio-demographic index (SDI), country, and territory. In addition, the burden of RHD was compared across age groups. From 1990 to 2019, the age-standardized incidence and prevalence rates of RHD increased by 14.4% (11.2–17.0%) and 13.8% (11.0–16.0%), respectively. Incidence and prevalence rates showed an increasing trend in low SDI and low-middle SDI locations, while high-middle SDI and high SDI locations showed a decreasing trend. The age-standardized DALYs and mortality rates of RHD decreased by 53.1% (46.4–60.0) and 56.9% (49.8–64.7%), and this downward trend was more prominent in high-middle SDI and middle SDI locations. In addition, the age of incidence and prevalence rates were concentrated between 5–24 years and 15–49 years, predominantly in poor regions, and RHD appeared to be more common in women than in men. Conclusion: The burden of RHD is negatively correlated with socioeconomic development status. In particular, the burden of RHD among children, adolescents, and women of childbearing age in poorer regions requires more attention. Policymakers should use the 2019 GBD data to guide cost-effective interventions and resource allocation for RHD.